In [35]:
import os
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline
sns.set_context('notebook', font_scale=1.5, rc={'lines.markeredgewidth': 2})
sns.set_style('white', {'axes.spines.right': False, 'axes.spines.top': False, 'xtick.bottom': True, 'ytick.left': True,})
sns.set_palette('deep')
In [36]:
%load_ext autoreload
%autoreload 2
In [37]:
import openscope_predictive_coding as opc
In [38]:
from openscope_predictive_coding.ophys.dataset.openscope_predictive_coding_dataset import OpenScopePredictiveCodingDataset
from openscope_predictive_coding.ophys.response_analysis.response_analysis import ResponseAnalysis
In [39]:
import openscope_predictive_coding.ophys.plotting.experiment_summary_figures as esf
import openscope_predictive_coding.ophys.plotting.summary_figures as sf
import openscope_predictive_coding.ophys.response_analysis.utilities as ut
In [40]:
cache_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\opc\opc_analysis'
In [193]:
manifest_file = os.path.join(cache_dir, 'opc_production_manifest.xlsx')
In [194]:
data = pd.read_excel(manifest_file)
In [ ]:
In [195]:
data = data[data['experiment_state']=='passed']
In [196]:
print('total mice:', len(data['mouse_id'].unique()))
print('# V1 injections:', len(data[data['injection_area']=='VISp']['mouse_id'].unique()))
print('# RSP injections:', len(data[data['injection_area']=='RSP']['mouse_id'].unique()))
print('# V1 imaging sessions:', len(data[data['imaging_area']=='VISp']['experiment_id'].unique()))
print('# RSP imaging sessions:', len(data[data['imaging_area']=='RSP']['experiment_id'].unique()))
print('# PM imaging sessions:', len(data[data['imaging_area']=='VISpm']['experiment_id'].unique()))
In [197]:
expt_ids = data.experiment_id.unique()
expt_ids = np.sort([int(e) for e in expt_ids if np.isnan(e) == False])
print(expt_ids)
get example expt¶
In [198]:
dataset = OpenScopePredictiveCodingDataset(expt_ids[1], cache_dir)
analysis = ResponseAnalysis(dataset, preload_response_dfs=False)
In [199]:
sequence_images = analysis.get_sequence_images()
oddball_images = analysis.get_oddball_images()
images = analysis.get_image_ids()
In [200]:
sequence_colors = sns.color_palette("Blues", len(sequence_images))
oddball_colors = sns.color_palette("Reds", len(oddball_images))
load multi session dfs¶
In [201]:
cache_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\opc\opc_analysis'
In [202]:
oddball_df = pd.read_hdf(os.path.join(cache_dir, 'multi_session_summary_dfs', 'mean_oddball_df_events.h5'))
In [203]:
transitions_df = pd.read_hdf(os.path.join(cache_dir, 'multi_session_summary_dfs', 'mean_transition_control_df_events.h5'))
In [204]:
occlusion_df = pd.read_hdf(os.path.join(cache_dir, 'multi_session_summary_dfs', 'mean_occlusion_df_events.h5'))
In [205]:
pre_df = pd.read_hdf(os.path.join(cache_dir, 'multi_session_summary_dfs', 'mean_randomized_control_pre_df_events.h5'))
# post_df = pd.read_hdf(os.path.join(cache_dir, 'multi_session_summary_dfs', 'mean_randomized_control_post_df_events.h5'))
# nmdf = pd.read_hdf(os.path.join(cache_dir, 'multi_session_summary_dfs', 'mean_natural_movie_one_df.h5'))
In [206]:
odf = oddball_df.copy()
ocdf = occlusion_df.copy()
tdf = transitions_df.copy()
In [207]:
cell_specimen_ids = odf.cell_specimen_id.unique()
merge with injection target¶
In [ ]:
def get_manifest(cache_dir=None):
"""
Loads experiment manifest file as a dataframe, listing all experiments and associated metadata
"""
if cache_dir is None:
cache_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\opc\opc_analysis'
manifest_file = os.path.join(cache_dir, 'opc_production_manifest.xlsx')
manifest = pd.read_excel(manifest_file)
return manifest
In [227]:
def add_projection_pathway_to_df(df, cache_dir=None):
"""
df: dataframe to add projection pathway information to. Dataframe must have a column 'experiment_id'.
cache_dir: cache directory to load manifest from
Adds columns called 'injection_area', the brain area where the retrograde tracer was injected,
and 'projection_pathway', indicating whether retrogradely labeled cells are part of a feed forward ('FF') or feed back ('FB') pathway,
to experiment manifest table, then merges in with provided dataframe.
"""
manifest = get_manifest(cache_dir)
manifest['projection_pathway'] = np.nan
manifest.at[manifest[manifest.injection_area=='RSP'].index.values, 'projection_pathway'] = 'FF'
manifest.at[manifest[manifest.injection_area=='VISp'].index.values, 'projection_pathway'] = 'FB'
manifest.at[manifest[(manifest.imaging_area=='VISpm')&(manifest.injection_area=='RSP')].index.values, 'projection_pathway'] = 'FF'
manifest.at[manifest[(manifest.imaging_area=='VISpm')&(manifest.injection_area=='VISp')].index.values, 'projection_pathway'] = 'FB'
df = df.merge(manifest[['experiment_id','injection_area','pathway']], on='experiment_id')
return data
In [228]:
odf = odf.merge(data[['experiment_id','injection_area','pathway']], on='experiment_id')
ocdf = ocdf.merge(data[['experiment_id','injection_area','pathway']], on='experiment_id')
tdf = tdf.merge(data[['experiment_id','injection_area','pathway']], on='experiment_id')
pre_df = pre_df.merge(data[['experiment_id','injection_area','pathway']], on='experiment_id')
annotate dfs¶
In [59]:
def add_location_to_df(df):
"""
Add useful columns, including 'depth' which translates 'imaging_depth' integer values in um to a string indicating superficial or deep layers,
and 'location', a string combining the imaged area and the 'depth' string as a way to easily group data by both area and depth for analysis.
"""
df['area'] = df.targeted_structure.values
df['depth'] = ['deep' if depth > 250 else 'superficial' for depth in df.imaging_depth.values]
df['location'] = None
df['location'] = [df.iloc[row].area+'_'+df.iloc[row].depth for row in range(len(df))]
return df
In [60]:
# odf = add_location_to_df(odf)
# ocdf = add_location_to_df(ocdf)
# tdf = add_location_to_df(tdf)
In [61]:
# pre_df = add_location_to_df(pre_df)
# post_df = add_location_to_df(post_df)
# nmdf = add_location_to_df(nmdf)
In [62]:
def add_retrogradely_labeled_column_to_df(df):
"""
takes any dataframe with a column for 'cell_specimen_id' and adds a new column called 'retrogradely_labeled' which is a boolean for whether the cell was tagged or not
"""
cache_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\opc\opc_analysis'
cache_dir = r'C:\Users\marinag\Dropbox\opc_analysis'
red_label_df = pd.read_hdf(os.path.join(cache_dir, 'multi_session_summary_dfs', 'red_label_df.h5'), key='df')
red_label_df.reset_index(inplace=True)
red_label_df['cell_specimen_id'] = [int(cell_specimen_id) for cell_specimen_id in red_label_df.cell_specimen_id.values]
red_df = red_label_df[['cell_specimen_id', 'retrogradely_labeled']]
df = pd.merge(df, red_df, left_on='cell_specimen_id', right_on='cell_specimen_id')
return df
In [63]:
# odf = add_retrogradely_labeled_column_to_df(odf)
# ocdf = add_retrogradely_labeled_column_to_df(ocdf)
# tdf = add_retrogradely_labeled_column_to_df(tdf)
In [64]:
# pre_df = add_retrogradely_labeled_column_to_df(pre_df)
# # post_df = add_retrogradely_labeled_column_to_df(post_df)
# nmdf = add_retrogradely_labeled_column_to_df(nmdf)
In [65]:
odf.keys()
Out[65]:
validate fraction responsive¶
In [66]:
def plot_mean_trace_with_variability(traces, frame_rate, ylabel='dF/F', label=None, color='k', interval_sec=1,
xlims=[-4, 4], ax=None):
# xlim = [xlims[0] + np.abs(xlims[0]), xlims[1] + np.abs(xlims[0])]
if ax is None:
fig, ax = plt.subplots()
if len(traces) > 0:
mean_trace = np.mean(traces, axis=0)
times = np.arange(0, len(mean_trace), 1)
sem = (traces.std()) / np.sqrt(float(len(traces)))
for trace in traces:
ax.plot(trace, linewidth=1, color='gray')
ax.plot(mean_trace, label=label, linewidth=3, color=color, zorder=100)
xticks, xticklabels = sf.get_xticks_xticklabels(mean_trace, frame_rate, interval_sec, window=xlims)
ax.set_xticks(xticks)
ax.set_xticklabels([int(x) for x in xticklabels])
ax.set_xlim(0,(np.abs(xlims[0])+xlims[1]) * int(frame_rate))
ax.set_xlabel('time (sec)')
ax.set_ylabel(ylabel)
sns.despine(ax=ax)
return ax
In [261]:
dataset = OpenScopePredictiveCodingDataset(expt_ids[1], cache_dir)
analysis = ResponseAnalysis(dataset, regenerate_dfs=True, use_events=True)
In [282]:
tmp = analysis.get_response_df('oddball')
In [283]:
tmp.keys()
Out[283]:
plot for one¶
In [79]:
save_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\opc\opc_analysis\summary_figures'
image_ids = sequence_images+oddball_images
experiment_id = expt_ids[0]
print(experiment_id)
dataset = OpenScopePredictiveCodingDataset(experiment_id, cache_dir)
analysis = ResponseAnalysis(dataset, preload_response_dfs=False, use_events=True)
oddball_df = analysis.get_response_df('oddball')
transition_df = analysis.get_response_df('transition_control')
cell_specimen_ids = dataset.get_cell_specimen_ids()
meta = dataset.metadata
plot mean response¶
In [95]:
cell_specimen_id = cell_specimen_ids[10]
figsize = (25,6)
fig, ax = plt.subplots(2, len(image_ids), figsize=figsize, sharey=True)
ax = ax.ravel()
for i,image_id in enumerate(image_ids):
df = oddball_df[(oddball_df.cell_specimen_id==cell_specimen_id)&(oddball_df.image_id==image_id)]
ax[i] = sf.plot_mean_trace(df.trace.values, 31., ylabel='dF/F', legend_label=None, color='k', interval_sec=0.5,
xlims=[-0.5, 0.5], ax=ax[i])
fraction_sig_trials = odf[(odf.cell_specimen_id==cell_specimen_id)&(odf.image_id==image_id)].fraction_significant_trials.values[0]
ax[i].set_title(str(int(image_id))+', '+str(np.round(fraction_sig_trials,2)))
ax[i].set_ylabel('')
ax[i].axvline(x=2*31, ymin=0, ymax=1, color=sns.color_palette()[0], linestyle='--')
df = transition_df[(transition_df.cell_specimen_id==cell_specimen_id)&(transition_df.image_id==image_id)&(transition_df.second_in_sequence==True)]
ax[i+len(image_ids)] = sf.plot_mean_trace(df.trace.values, 31., ylabel='dF/F', legend_label=None, color='k', interval_sec=0.5,
xlims=[-0.5, 0.5], ax=ax[i+len(image_ids)])
fraction_sig_trials = tdf[(tdf.cell_specimen_id==cell_specimen_id)&(tdf.image_id==image_id)].fraction_significant_trials.values[0]
ax[i+len(image_ids)].set_title(str(int(image_id))+', '+str(np.round(fraction_sig_trials,2)))
ax[i+len(image_ids)].set_ylabel('')
ax[i+len(image_ids)].axvline(x=2*31, ymin=0, ymax=1, color=sns.color_palette()[0], linestyle='--')
ax[0].set_ylabel('dF/F')
ax[len(image_ids)].set_ylabel('dF/F')
fig.tight_layout()
plt.gcf().subplots_adjust(top=0.85)
title = 'cell_specimen_id: '+str(cell_specimen_id)+', experiment_id: '+str(meta.ophys_experiment_id.values[0])+', area: '+str(meta.targeted_structure.values[0]+', depth: '+str(meta.imaging_depth.values[0]))
plt.suptitle(title, x=0.5, y=1.0, horizontalalignment='center')
fig_title = str(meta.ophys_experiment_id.values[0])+'_'+str(cell_specimen_id)+'_'+str(meta.targeted_structure.values[0]+'_'+str(meta.imaging_depth.values[0]))
sf.save_figure(fig, figsize, save_dir, 'mean_response_oddball_transition_events', str(cell_specimen_id))
# plt.close()
In [81]:
plt.plot(oddball_df.trace.mean())
Out[81]:
In [82]:
plt.plot(transition_df.trace.mean())
Out[82]:
plot for all¶
In [ ]:
# save_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\opc\opc_analysis\summary_figures'
# image_ids = sequence_images+oddball_images
# for experiment_id in expt_ids:
# print(experiment_id)
# dataset = OpenScopePredictiveCodingDataset(experiment_id, cache_dir)
# analysis = ResponseAnalysis(dataset, preload_response_dfs=False)
# oddball_df = analysis.get_response_df('oddball')
# transition_df = analysis.get_response_df('transition_control')
# cell_specimen_ids = dataset.get_cell_specimen_ids()
# meta = dataset.metadata
# for cell_specimen_id in cell_specimen_ids:
# figsize = (25,6)
# fig, ax = plt.subplots(2, len(image_ids), figsize=figsize, sharey=True)
# ax = ax.ravel()
# for i,image_id in enumerate(image_ids):
# df = oddball_df[(oddball_df.cell_specimen_id==cell_specimen_id)&(oddball_df.image_id==image_id)]
# ax[i] = sf.plot_mean_trace(df.dff_trace.values, 31., ylabel='dF/F', legend_label=None, color='k', interval_sec=1,
# xlims=[-2, 2], ax=ax[i])
# fraction_sig_trials = odf[(odf.cell_specimen_id==cell_specimen_id)&(odf.image_id==image_id)].fraction_significant_trials.values[0]
# ax[i].set_title(str(int(image_id))+', '+str(np.round(fraction_sig_trials,2)))
# ax[i].set_ylabel('')
# ax[i].axvline(x=2*31, ymin=0, ymax=1, color=sns.color_palette()[0], linestyle='--')
# df = transition_df[(transition_df.cell_specimen_id==cell_specimen_id)&(transition_df.image_id==image_id)&(transition_df.second_in_sequence==True)]
# ax[i+len(image_ids)] = sf.plot_mean_trace(df.dff_trace.values, 31., ylabel='dF/F', legend_label=None, color='k', interval_sec=1,
# xlims=[-2, 2], ax=ax[i+len(image_ids)])
# fraction_sig_trials = tdf[(tdf.cell_specimen_id==cell_specimen_id)&(tdf.image_id==image_id)].fraction_significant_trials.values[0]
# ax[i+len(image_ids)].set_title(str(int(image_id))+', '+str(np.round(fraction_sig_trials,2)))
# ax[i+len(image_ids)].set_ylabel('')
# ax[i+len(image_ids)].axvline(x=2*31, ymin=0, ymax=1, color=sns.color_palette()[0], linestyle='--')
# ax[0].set_ylabel('dF/F')
# ax[len(image_ids)].set_ylabel('dF/F')
# fig.tight_layout()
# plt.gcf().subplots_adjust(top=0.85)
# title = 'cell_specimen_id: '+str(cell_specimen_id)+', experiment_id: '+str(meta.ophys_experiment_id.values[0])+', area: '+str(meta.targeted_structure.values[0]+', depth: '+str(meta.imaging_depth.values[0]))
# plt.suptitle(title, x=0.5, y=1.0, horizontalalignment='center')
# fig_title = str(meta.ophys_experiment_id.values[0])+'_'+str(cell_specimen_id)+'_'+str(meta.targeted_structure.values[0]+'_'+str(meta.imaging_depth.values[0]))
# sf.save_figure(fig, figsize, save_dir, 'mean_response_oddball_transition', fig_title)
# plt.close()
plot variability¶
for one¶
In [100]:
cell_specimen_id = cell_specimen_ids[0]
figsize = (25,6)
fig, ax = plt.subplots(2, len(image_ids), figsize=figsize, sharey=True)
ax = ax.ravel()
for i,image_id in enumerate(image_ids):
df = oddball_df[(oddball_df.cell_specimen_id==cell_specimen_id)&(oddball_df.image_id==image_id)]
ax[i] = plot_mean_trace_with_variability(df.trace.values, 31., ylabel='dF/F', label=None, color='k', interval_sec=1,
xlims=[-0.5, 0.5], ax=ax[i])
fraction_sig_trials = odf[(odf.cell_specimen_id==cell_specimen_id)&(odf.image_id==image_id)].fraction_significant_trials.values[0]
ax[i].set_title(str(int(image_id))+', '+str(np.round(fraction_sig_trials,2)))
ax[i].set_ylabel('')
ax[i].axvline(x=2*31, ymin=0, ymax=1, color=sns.color_palette()[0], linestyle='--')
df = transition_df[(transition_df.cell_specimen_id==cell_specimen_id)&(transition_df.image_id==image_id)&(transition_df.second_in_sequence==True)]
ax[i+len(image_ids)] = plot_mean_trace_with_variability(df.trace.values, 31., ylabel='dF/F', label=None, color='k', interval_sec=1,
xlims=[-0.5, 0.5], ax=ax[i+len(image_ids)])
fraction_sig_trials = tdf[(tdf.cell_specimen_id==cell_specimen_id)&(tdf.image_id==image_id)].fraction_significant_trials.values[0]
ax[i+len(image_ids)].set_title(str(int(image_id))+', '+str(np.round(fraction_sig_trials,2)))
ax[i+len(image_ids)].set_ylabel('')
ax[i+len(image_ids)].axvline(x=2*31, ymin=0, ymax=1, color=sns.color_palette()[0], linestyle='--')
ax[0].set_ylabel('dF/F')
ax[len(image_ids)].set_ylabel('dF/F')
fig.tight_layout()
plt.gcf().subplots_adjust(top=0.85)
title = 'cell_specimen_id: '+str(cell_specimen_id)+', experiment_id: '+str(meta.ophys_experiment_id.values[0])+', area: '+str(meta.targeted_structure.values[0]+', depth: '+str(meta.imaging_depth.values[0]))
plt.suptitle(title, x=0.5, y=1.0, horizontalalignment='center')
fig_title = str(meta.ophys_experiment_id.values[0])+'_'+str(cell_specimen_id)+'_'+str(meta.targeted_structure.values[0])+'_'+str(meta.imaging_depth.values[0])
sf.save_figure(fig, figsize, save_dir, 'variability_events', fig_title)
# plt.close()
plot for all¶
In [ ]:
# save_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\opc\opc_analysis\summary_figures'
# image_ids = sequence_images+oddball_images
# for experiment_id in expt_ids:
# print(experiment_id)
# dataset = OpenScopePredictiveCodingDataset(experiment_id, cache_dir)
# analysis = ResponseAnalysis(dataset, preload_response_dfs=False)
# oddball_df = analysis.get_response_df('oddball')
# transition_df = analysis.get_response_df('transition_control')
# cell_specimen_ids = dataset.get_cell_specimen_ids()
# meta = dataset.metadata
# for cell_specimen_id in cell_specimen_ids:
# figsize = (25,6)
# fig, ax = plt.subplots(2, len(image_ids), figsize=figsize, sharey=True)
# ax = ax.ravel()
# for i,image_id in enumerate(image_ids):
# df = oddball_df[(oddball_df.cell_specimen_id==cell_specimen_id)&(oddball_df.image_id==image_id)]
# ax[i] = plot_mean_trace_with_variability(df.dff_trace.values, 31., ylabel='dF/F', label=None, color='k', interval_sec=1,
# xlims=[-2, 2], ax=ax[i])
# fraction_sig_trials = odf[(odf.cell_specimen_id==cell_specimen_id)&(odf.image_id==image_id)].fraction_significant_trials.values[0]
# ax[i].set_title(str(int(image_id))+', '+str(np.round(fraction_sig_trials,2)))
# ax[i].set_ylabel('')
# ax[i].axvline(x=2*31, ymin=0, ymax=1, color=sns.color_palette()[0], linestyle='--')
# df = transition_df[(transition_df.cell_specimen_id==cell_specimen_id)&(transition_df.image_id==image_id)&(transition_df.second_in_sequence==True)]
# ax[i+len(image_ids)] = plot_mean_trace_with_variability(df.dff_trace.values, 31., ylabel='dF/F', label=None, color='k', interval_sec=1,
# xlims=[-2, 2], ax=ax[i+len(image_ids)])
# fraction_sig_trials = tdf[(tdf.cell_specimen_id==cell_specimen_id)&(tdf.image_id==image_id)].fraction_significant_trials.values[0]
# ax[i+len(image_ids)].set_title(str(int(image_id))+', '+str(np.round(fraction_sig_trials,2)))
# ax[i+len(image_ids)].set_ylabel('')
# ax[i+len(image_ids)].axvline(x=2*31, ymin=0, ymax=1, color=sns.color_palette()[0], linestyle='--')
# ax[0].set_ylabel('dF/F')
# ax[len(image_ids)].set_ylabel('dF/F')
# fig.tight_layout()
# plt.gcf().subplots_adjust(top=0.85)
# title = 'cell_specimen_id: '+str(cell_specimen_id)+', experiment_id: '+str(meta.ophys_experiment_id.values[0])+', area: '+str(meta.targeted_structure.values[0]+', depth: '+str(meta.imaging_depth.values[0]))
# plt.suptitle(title, x=0.5, y=1.0, horizontalalignment='center')
# fig_title = str(meta.ophys_experiment_id.values[0])+'_'+str(cell_specimen_id)+'_'+str(meta.targeted_structure.values[0]+'_'+str(meta.imaging_depth.values[0])
# sf.save_figure(fig, figsize, save_dir, 'variability', fig_title)
# plt.close()
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
population averages for different conditions¶
In [217]:
def get_xticks_xticklabels(trace, frame_rate, interval_sec=1, window=[-2,2]):
"""
Function that accepts a timeseries, evaluates the number of points in the trace, and converts from acquisition frames to timestamps
:param trace: a single trace where length = the number of timepoints
:param frame_rate: ophys frame rate if plotting a calcium trace, stimulus frame rate if plotting running speed
:param interval_sec: interval in seconds in between labels
:return: xticks, xticklabels = xticks in frames corresponding to timepoints in the trace, xticklabels in seconds
"""
interval_frames = interval_sec * frame_rate
n_frames = len(trace)
n_sec = n_frames / frame_rate
xticks = np.arange(0, n_frames + 5, interval_frames)
xticklabels = np.arange(0, n_sec + 0.1, interval_sec)
xticklabels = xticklabels + window[0]
if interval_sec >= 1:
xticklabels = [int(x) for x in xticklabels]
return xticks, xticklabels
In [218]:
def plot_response_across_conditions_population(df, condition='area', conditions=['VISp', 'VISpm', 'RSP'],
window=[-2, 2], save_figures=False, colors=None, autoscale=False,
title='',save_dir=None, folder=None, ax=None, pref_stim=False, frame_rate=31.):
if pref_stim:
df = df[df.pref_stim == True].copy()
if ax is None:
figsize = (5, 5)
fig, ax = plt.subplots(figsize=figsize)
for c, condition_value in enumerate(conditions):
tmp = df[df[condition] == condition_value]
if colors is None:
colors = sns.color_palette()
traces = tmp.mean_trace.values
trace = np.mean(traces, axis=0)
ax = sf.plot_mean_trace(traces, frame_rate, legend_label=condition_value, color=colors[c],
interval_sec=0.5, xlims=window, ax=ax)
ax.axvspan(np.abs(window[0])*frame_rate, (np.abs(window[0])+0.25)*frame_rate,
facecolor='gray', edgecolor='none', alpha=0.7, linewidth=0, zorder=1)
# ax = plot_flashes_on_trace(ax, flashes=False, alpha=0.15, window=window, omitted=True)
xticks, xticklabels = get_xticks_xticklabels(trace, 31., interval_sec=1, window=window)
ax.set_xticks(xticks)
ax.set_xticklabels([int(x) for x in xticklabels])
# ax.set_xlim(0,(np.abs(window[0])+window[1])*31.)
ax.legend(bbox_to_anchor=(1.1, 1), title=condition)
if not autoscale:
ymin, ymax = ax.get_ylim()
if ymin > 0:
ax.set_ylim(0, ymax * 1.2)
else:
ax.set_ylim(ymin * 1.2, ymax * 1.2)
ax.set_title(title)
if save_figures:
fig.tight_layout()
save_figure(fig, figsize, save_dir, folder, 'population_response_across_' + condition + '_' + cre_line)
return ax
In [219]:
areas = ['VISp','VISpm','RSP']
In [220]:
odf['expected'] = np.nan
In [221]:
expected_image = analysis.get_sequence_images()[3]
oddball_images = analysis.get_oddball_images()
In [222]:
expected_list = []
for image_id in odf.image_id.values:
if image_id == expected_image:
expected_list.append(True)
elif image_id in oddball_images:
expected_list.append(False)
else:
expected_list.append(np.nan)
odf['expected'] = expected_list
In [223]:
odf.expected.unique()
Out[223]:
In [108]:
odf[['image_id', 'oddball', 'expected']]
Out[108]:
In [109]:
experiment_id = 828956958
fig, ax = plt.subplots()
df = odf[(odf.experiment_id==experiment_id)].copy()
plot_response_across_conditions_population(df, condition='expected', conditions=[True, False],
window=[-2, 2], ax=ax)
Out[109]:
In [110]:
c = sns.color_palette()
colors = [c[0], c[3]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)].copy()
ax[i] = plot_response_across_conditions_population(df, condition='expected', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
In [111]:
c = sns.color_palette()
colors = [c[0], c[3]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.retrogradely_labeled==True)].copy()
ax[i] = plot_response_across_conditions_population(df, condition='expected', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('retrogradely labeled')
Out[111]:
In [112]:
c = sns.color_palette()
colors = [c[0], c[3]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.retrogradely_labeled==False)].copy()
ax[i] = plot_response_across_conditions_population(df, condition='expected', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('not retrogradely labeled')
Out[112]:
In [1]:
c = sns.color_palette()
colors = [c[3], c[2]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.oddball==True)].copy()
ax[i] = plot_response_across_conditions_population(df, condition='retrogradely_labeled', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('oddball response')
In [130]:
c = sns.color_palette()
colors = [c[3], c[2]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.oddball==False)&(odf.image_id==sequence_images[0])].copy()
ax[i] = plot_response_across_conditions_population(df, condition='retrogradely_labeled', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('first image in sequence - image A')
Out[130]:
In [131]:
c = sns.color_palette()
colors = [c[3], c[2]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.image_id==sequence_images[3])].copy()
ax[i] = plot_response_across_conditions_population(df, condition='retrogradely_labeled', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('expected sequence image - image D')
Out[131]:
In [231]:
odf.pathway.unique()
Out[231]:
In [241]:
c = sns.color_palette()
colors = [c[3], c[2]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.oddball==True)&(odf.retrogradely_labeled==True)].copy()
ax[i] = plot_response_across_conditions_population(df, condition='pathway', conditions=df.pathway.unique(),
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('oddball images - retrogradely labeled')
Out[241]:
In [242]:
c = sns.color_palette()
colors = [c[3], c[2]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.oddball==True)&(odf.retrogradely_labeled==False)].copy()
ax[i] = plot_response_across_conditions_population(df, condition='pathway', conditions=df.pathway.unique(),
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('oddball images - not labeled')
Out[242]:
In [239]:
c = sns.color_palette()
colors = [c[3], c[2]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.image_id==sequence_images[3])].copy()
ax[i] = plot_response_across_conditions_population(df, condition='pathway', conditions=df.pathway.unique(),
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('expected sequence image - image D')
Out[239]:
In [ ]:
In [ ]:
In [ ]:
single cells¶
In [137]:
from openscope_predictive_coding.ophys.plotting import population_summary_figures as psf
In [ ]:
c = sns.color_palette()
colors = [c[3], c[2]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.oddball==False)&(odf.image_id==sequence_images[0])&()].copy()
ax[i] = plot_response_across_conditions_population(df, condition='retrogradely_labeled', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('first image in sequence - image A')
In [149]:
cdf.keys()
Out[149]:
In [157]:
for cell_specimen_id in odf[(odf.targeted_structure=='RSP')&(odf.experiment_id==830688102)].cell_specimen_id.unique():
figsize = (20,4)
fig, ax = plt.subplots(1,5, figsize=figsize, sharey=True)
ax = ax.ravel()
cdf = odf[odf.cell_specimen_id==cell_specimen_id]
for i, sequence_image in enumerate(sequence_images):
df = cdf[(cdf.image_id==sequence_image)].copy()
psf.plot_mean_trace_from_mean_df(df, ax=ax[i], xlims=[-2,2])
ax[i].axvline(x=31*2, linestyle='--')
ax[i].set_title(sequence_image)
i+=1
df = cdf[(cdf.oddball==True)].copy()
psf.plot_mean_trace_from_mean_df(df, ax=ax[i], xlims=[-2,2])
ax[i].set_title('oddball')
ax[i].axvline(x=31*2, linestyle='--')
plt.suptitle(str(cdf.cell_specimen_id.values[0])+'_'+cdf.targeted_structure.values[0]+'_'+str(cdf.retrogradely_labeled.values[0]),x=0.5, y=1.05)
fig.tight_layout()
In [158]:
for cell_specimen_id in odf[(odf.targeted_structure=='VISp')].cell_specimen_id.unique()[:100]:
figsize = (20,4)
fig, ax = plt.subplots(1,5, figsize=figsize, sharey=True)
ax = ax.ravel()
cdf = odf[odf.cell_specimen_id==cell_specimen_id]
for i, sequence_image in enumerate(sequence_images):
df = cdf[(cdf.image_id==sequence_image)].copy()
psf.plot_mean_trace_from_mean_df(df, ax=ax[i], xlims=[-2,2])
ax[i].axvline(x=31*2, linestyle='--')
ax[i].set_title(sequence_image)
i+=1
df = cdf[(cdf.oddball==True)].copy()
psf.plot_mean_trace_from_mean_df(df, ax=ax[i], xlims=[-2,2])
ax[i].set_title('oddball')
ax[i].axvline(x=31*2, linestyle='--')
plt.suptitle(str(cdf.cell_specimen_id.values[0])+'_'+cdf.targeted_structure.values[0]+'_'+str(cdf.retrogradely_labeled.values[0]),x=0.5, y=1.05)
fig.tight_layout()
In [ ]:
In [ ]:
heatmaps¶
In [323]:
def reannotate_pref_stim(df):
m = df.copy()
m['pref_stim'] = False
for cell in m.cell_specimen_id.values:
cdf = m[m.cell_specimen_id==cell]
image_index = np.where(cdf['mean_response'].values == np.max(cdf['mean_response'].values))[0][0]
pref_image = cdf['mean_response'].index[image_index]
rows = df[(df.cell_specimen_id == cell) & (df.image_id == pref_image)].index
for trial in trials:
df.loc[rows, 'pref_stim'] = True
return df
In [326]:
def plot_mean_image_responses(df, image_ids, vmax=0.03, colorbar=False, ax=None, save_dir=None, folder=None, use_events=True,
interval_sec=1, window=[-2,2]):
if use_events:
# vmax = 0.003
label = 'mean event magnitude'
suffix = '_events'
else:
# vmax = 0.3
label = 'mean dF/F'
suffix = ''
experiment_id = df.experiment_id.unique()[0]
imaging_area = df.area.unique()[0]
df = reannotate_pref_stim(df[df.image_id.isin(image_ids)])
figsize = (25,10)
cells = []
for image_id in image_ids:
tmp = df[(df.image_id == image_id) & (df.pref_stim == True)]
order = np.argsort(tmp.mean_response.values)[::-1]
cell_ids = list(tmp.cell_specimen_id.values[order])
cells = cells + cell_ids
if ax is None:
fig, ax = plt.subplots(1, len(image_ids), figsize=figsize, sharey=True, sharex=True)
ax = ax.ravel()
for i, image_id in enumerate(image_ids):
im_df = df[(df.image_id == image_id)]
len_trace = len(im_df.mean_trace.values[0])
response_array = np.empty((len(cells), len_trace))
for x, cell in enumerate(cells):
tmp = im_df[im_df.cell_specimen_id == cell]
if len(tmp) >= 1:
trace = tmp.mean_trace.values[0]
else:
trace = np.empty((len_trace))
trace[:] = np.nan
response_array[x, :] = trace
sns.heatmap(data=response_array, vmin=0, vmax=vmax, ax=ax[i], cmap='magma', cbar=colorbar,
cbar_kws={'label': label})
xticks, xticklabels = sf.get_xticks_xticklabels(trace, 31., interval_sec=interval_sec, window=window)
ax[i].set_xticks(xticks)
if interval_sec < 1:
ax[i].set_xticklabels(xticklabels)
else:
ax[i].set_xticklabels([int(x) for x in xticklabels])
if response_array.shape[0] > 300:
interval = 500
else:
interval = 50
ax[i].set_xlim(0, (np.abs(window[0])+window[1])*31.)
ax[i].set_yticks(np.arange(0, response_array.shape[0], interval))
ax[i].set_yticklabels(np.arange(0, response_array.shape[0], interval))
ax[i].set_xlabel('time (sec)', fontsize=16)
ax[i].set_title(str(int(image_id)))
ax[0].set_ylabel('cells')
plt.suptitle(str(experiment_id)+', '+imaging_area, x=0.52, y=.98, fontsize=18, horizontalalignment='center')
fig.tight_layout()
plt.gcf().subplots_adjust(top=0.85)
# if save_dir:
# save_figure(fig, figsize, save_dir, folder,
# 'change_response_matrix_' + cre_line + '_' + image_set + '_' + suffix)
In [ ]:
In [ ]:
In [316]:
image_ids = sequence_images+oddball_images
plot_mean_image_responses(odf, image_ids, vmax=0.005, colorbar=False, ax=None, save_dir=None, folder=None, use_events=True,
interval_sec=1, window=[-2,2])
In [327]:
plot_mean_image_responses(odf[odf.area==area], image_ids, vmax=0.008, colorbar=False, ax=None, save_dir=None, folder=None, use_events=True,
interval_sec=1, window=[-2,2])
In [322]:
image_ids = sequence_images
for area in odf.area.unique():
plot_mean_image_responses(odf[odf.area==area], image_ids, vmax=0.008, colorbar=False, ax=None, save_dir=None, folder=None, use_events=True,
interval_sec=1, window=[-2,2])
In [321]:
image_ids = sequence_images
for experiment_id in odf.experiment_id.unique():
try:
plot_mean_image_responses(odf[odf.experiment_id==experiment_id], image_ids, vmax=0.008, colorbar=False, ax=None, save_dir=None, folder=None, use_events=True,
interval_sec=1, window=[-2,2])
except:
print(experiment_id)
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [133]:
c = sns.color_palette()
colors = [c[0], c[3]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.depth=='superficial')].copy()
ax[i] = plot_response_across_conditions_population(df, condition='expected', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('superficial')
Out[133]:
In [134]:
c = sns.color_palette()
colors = [c[0], c[3]]
figsize = (15,5)
fig, ax = plt.subplots(1,3, figsize=figsize, sharey=True)
for i,area in enumerate(areas):
df = odf[(odf.area==area)&(odf.depth=='deep')].copy()
ax[i] = plot_response_across_conditions_population(df, condition='expected', conditions=[True, False],
window=[-2, 2], ax=ax[i], colors=colors, title=area)
plt.suptitle('deep')
Out[134]:
In [132]:
odf
Out[132]:
oddball response by session¶
In [135]:
figsize=(20,20)
fig, ax = plt.subplots(7,8,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
i=0
for experiment_id in odf.experiment_id.unique():
edf = odf[odf.experiment_id==experiment_id]
sequence = edf[edf.image_id==images[3]].mean_trace.mean()
oddballs = edf[edf.oddball==True].mean_trace.mean()
ax[i].plot(sequence, color='blue', label='sequence')
ax[i].plot(oddballs, color='red', label='oddballs')
xticks, xticklabels = sf.get_xticks_xticklabels(sequence, 31., interval_sec=1)
ax[i].set_xticks(xticks)
ax[i].set_xticklabels([int(x) for x in xticklabels])
ax[i].set_ylabel('dF/F')
ax[i].set_xlabel('time (sec)')
ax[i].set_title(str(experiment_id)+'_'+str(edf.donor_id.values[0])+'\n'+edf.targeted_structure.values[0]+', '+str(edf.imaging_depth.values[0]))
i+=1
fig.tight_layout()
In [ ]:
In [ ]:
oddball response metric - transitions control¶
In [104]:
odf = odf.reset_index()
oddball_index = []
for index in odf.index.values:
row_df = odf.iloc[index]
cell_specimen_id = row_df.cell_specimen_id
image_id = row_df.image_id
oddball_mean_response = row_df.mean_response
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)&(tdf.second_in_sequence==True)]
transitions_mean_response = tcdf.mean_response.values[0]
oddball_response_index = ((oddball_mean_response-transitions_mean_response)/(oddball_mean_response+transitions_mean_response))
oddball_index.append(oddball_response_index)
odf['oddball_response_index'] = oddball_index
In [43]:
def plot_oddball_transition_responses(odf, tdf, cell_specimen_id, save_dir=None):
images = oddball_images+sequence_images
figsize=(20,2.5)
fig, ax = plt.subplots(1, len(images), figsize=figsize, sharey=True)
for i,image_id in enumerate(images):
cell_odf = odf[(odf.cell_specimen_id==cell_specimen_id)&(odf.image_id==image_id)]
cell_tdf = tdf[(tdf.cell_specimen_id==cell_specimen_id)&(tdf.image_id==image_id)]
ax[i].plot(cell_odf.mean_trace.values[0], color=sns.color_palette()[3], label='oddball_block')
ax[i].plot(cell_tdf.mean_trace.values[0], color=sns.color_palette()[0], label='transition_control')
xticks, xticklabels = sf.get_xticks_xticklabels(cell_odf.mean_trace.values[0], 31.)
ax[i].set_xticks(xticks)
ax[i].set_xticklabels(xticklabels)
ax[i].set_xlabel('time (s)')
ax[i].set_title(str(np.round(cell_odf.oddball_response_index.values[0], 2))+', '+str(np.round(cell_odf.fraction_significant_trials.values[0], 2)),fontsize=12)
ax[0].set_ylabel('dF/F')
ax[i].legend(fontsize='small')
fig.tight_layout()
fig.subplots_adjust(bottom=0.25)
if save_dir:
sf.save_figure(fig, figsize, save_dir, 'oddball_transition_responses', str(int(cell_specimen_id)))
plt.close()
In [44]:
cell_specimen_id = odf.cell_specimen_id.unique()[0]
plot_oddball_transition_responses(odf, tdf, cell_specimen_id, save_dir=None)
In [45]:
# for cell_specimen_id in odf.cell_specimen_id.unique():
# plot_oddball_transition_responses(odf, tdf, cell_specimen_id, save_dir=save_dir)
In [151]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = odf.copy()
df = df[df.oddball==True]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
df = df[df.fraction_significant_trials>0.1]
colors = [sns.color_palette()[0], sns.color_palette()[3]]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='oddball_response_index', order=locations,
hue='retrogradely_labeled', palette=colors, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
ax.set_ylim(-0.02, 0.8)
legend = ax.legend(fontsize='x-small', title='retrogradely labeled', loc='upper right')
plt.setp(legend.get_title(),fontsize='x-small')
Out[151]:
In [155]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
metric = 'oddball_response_index'
df = odf.copy()
df = df[df.oddball==True]
df = df[(df[metric]>-5)&(df[metric]<5)]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
df = df[df.fraction_significant_trials>0.01]
colors = [sns.color_palette()[0], sns.color_palette()[3], sns.color_palette()[2]]
fig, ax = plt.subplots(1, figsize=(5,5), sharey=True, sharex=True)
for i, area in enumerate(['VISp', 'VISpm', 'RSP']):
ax = sns.distplot(df[(df.area==area)&(df.retrogradely_labeled==label)][metric].values, bins=30,
color=colors[i], ax=ax, label=area)
ax.set_xlabel(metric)
legend = ax.legend(fontsize='x-small', title='area', loc='upper right')
plt.setp(legend.get_title(),fontsize='x-small')
ax.set_ylabel('density')
# ax.set_title(area)
In [159]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
metric = 'oddball_response_index'
df = odf.copy()
df = df[df.oddball==True]
df = df[(df[metric]>-5)&(df[metric]<5)]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
df = df[df.fraction_significant_trials>0.01]
colors = [sns.color_palette()[0], sns.color_palette()[3]]
fig, ax = plt.subplots(1,3, figsize=(15,5), sharey=True, sharex=True)
for i, area in enumerate(['VISp', 'VISpm', 'RSP']):
for c, label in enumerate(df.retrogradely_labeled.unique()):
ax[i] = sns.distplot(df[(df.area==area)&(df.retrogradely_labeled==label)][metric].values, bins=30,
color=colors[c], ax=ax[i], label=label)
ax[i].set_xlabel(metric)
legend = ax[i].legend(fontsize='x-small', title='retrogradely labeled', loc='upper right')
plt.setp(legend.get_title(),fontsize='x-small')
ax[i].set_title(area)
ax[0].set_ylabel('density')
Out[159]:
In [158]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
metric = 'oddball_response_index'
df = odf.copy()
df = df[df.oddball==True]
df = df[(df[metric]>-5)&(df[metric]<5)]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
df = df[df.fraction_significant_trials>0.1]
fig, ax = plt.subplots(1,3, figsize=(15,5), sharey=True, sharex=True)
for i, area in enumerate(['VISp', 'VISpm', 'RSP']):
for c, depth in enumerate(df.depth.unique()):
ax[i] = sns.distplot(df[(df.area==area)&(df.depth==depth)][metric].values, bins=30,
color=colors[c], ax=ax[i], label=depth)
ax[i].set_xlabel(metric)
ax[i].legend()
ax[i].set_title(area)
ax[0].set_ylabel('density')
Out[158]:
In [49]:
metric = 'oddball_response_index'
df = odf.copy()
df = df[df.oddball==True]
df = df[(df[metric]>-5)&(df[metric]<5)]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots(1,2, figsize=(10,5), sharey=True, sharex=True)
for i, depth in enumerate(df.depth.unique()):
for c, label in enumerate(df.retrogradely_labeled.unique()):
ax[i] = sns.distplot(df[(df.depth==depth)&(df.retrogradely_labeled==label)][metric].values, bins=30,
color=colors[c], ax=ax[i], label=label)
ax[i].set_xlabel(metric)
ax[i].legend()
ax[i].set_title(depth)
In [50]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = odf.copy()
df = df[df.oddball==True]
# df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots()
ax = sns.distplot(data=df, x='location', y='oddball_response_index', order=locations, hue='retrogradely_labeled', ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
oddball response metric - randomized control¶
In [ ]:
odf = odf.reset_index()
oddball_index = []
for index in odf.index.values:
row_df = odf.iloc[index]
cell_specimen_id = row_df.cell_specimen_id
image_id = row_df.image_id
oddball_mean_response = row_df.mean_response
pre_cdf = pre_df[(pre_df.image_id==image_id)&(pre_df.cell_specimen_id==cell_specimen_id)]
control_mean_response = pre_cdf.mean_response.values[0]
oddball_response_index = ((oddball_mean_response-control_mean_response)/(oddball_mean_response+control_mean_response))
oddball_index.append(oddball_response_index)
odf['oddball_response_index_control'] = oddball_index
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
occlusion metric¶
In [174]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = ocdf.copy()
# df = df[df.oddball==True]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.1]
cells = df[df.fraction_significant_trials>0.1].cell_specimen_id.unique()
df = df[df.cell_specimen_id.isin(cells)]
colors = [sns.color_palette()[0], sns.color_palette()[3]]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='mean_response', order=locations, hue='fraction_occlusion',
palette=sns.color_palette('Blues'), ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
ax.legend()
# ax.set_ylim(0,0.8)
ax.set_title('occlusion population response')
Out[174]:
In [173]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = ocdf.copy()
# df = df[df.oddball==True]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
cells = df[df.fraction_significant_trials>0.1].cell_specimen_id.unique()
df = df[df.cell_specimen_id.isin(cells)]
colors = [sns.color_palette()[0], sns.color_palette()[3]]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='fraction_occlusion', y='mean_response', order=np.sort(df.fraction_occlusion.unique()),
hue='location', hue_order=locations, palette=sns.color_palette(), ax=ax)
ax.set_xticklabels(np.sort(df.fraction_occlusion.unique()))
ax.set_xlabel('fraction occlusion')
# ax.set_ylim(0,0.8)
ax.set_title('occlusion tuning')
legend = ax.legend(fontsize='x-small', title='location', bbox_to_anchor=(1,1))
plt.setp(legend.get_title(),fontsize='x-small')
Out[173]:
In [ ]:
In [166]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = ocdf.copy()
# df = df[df.oddball==True]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.1]
colors = [sns.color_palette()[0], sns.color_palette()[3]]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='mean_response', order=locations, hue='retrogradely_labeled',
palette=colors, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
ax.set_ylim(0,0.6)
legend = ax.legend(fontsize='x-small', title='retrogradely labeled', loc='upper right')
plt.setp(legend.get_title(),fontsize='x-small')
ax.set_title('occlusion population response')
Out[166]:
need to calculate the pref amount of occlusion¶
In [ ]:
In [ ]:
In [ ]:
# locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
# df = ocdf.copy()
# # df = df[df.oddball==True]
# # df = df[df.pref_stim==True]
# # df = df[df.mean_response>0.05]
# colors = sns.color_palette('Blues')
# fig, ax = plt.subplots()
# for i, location in enumerate(locations):
# tmp = df[df.location==location]
# for r, occ in enumerate(tmp.fraction_occlusion.unique()):
# fraction_pref = len(tmp[(tmp.pref_stim==True)&(tmp.fraction_occlusion==occ)])/float(len(tmp.cell_specimen_id.unique()))
# ax.plot(i, fraction_pref, 'o', color=colors[r])
# for r, occ in enumerate(tmp.fraction_occlusion.unique()):
# fraction_pref = len(tmp[(tmp.pref_stim==True)&(tmp.fraction_occlusion==occ)])/float(len(tmp.cell_specimen_id.unique()))
# ax.plot(i, fraction_pref, 'o', color=colors[r], label=occ)
# ax.set_xticks(range(len(locations)))
# ax.set_xticklabels(locations, rotation=90)
# ax.set_xlabel('')
# ax.set_ylim(-0.05,1.05)
# legend = ax.legend(fontsize='x-small', title='fraction occlusion', loc='upper right')
# plt.setp(legend.get_title(),fontsize='x-small')
# ax.set_ylabel('fraction responsive cells')
# ax.set_title('occlusion responsiveness');
In [ ]:
In [ ]:
In [213]:
def occlusion_response_index_byimage(dfc, val='mean_response'):
dfc_occlusion = dfc[dfc['fraction_occlusion']!=0]
dfc_noocclusion = dfc[dfc['fraction_occlusion']==0]
occlusion_zscore = abs(dfc_occlusion.groupby('image_id')[val].mean()-dfc_noocclusion.groupby('image_id')[val].mean())/dfc_occlusion.groupby('image_id')[val].std()
return occlusion_zscore
def occlusion_response_index_bycell(dfc, val='mean_response'):
# dfc = dfc[dfc.pref_stim==True]
dfc_occlusion = dfc[dfc['fraction_occlusion']!=0]
dfc_noocclusion = dfc[dfc['fraction_occlusion']==0]
occlusion_zscore = abs(dfc_occlusion.groupby(['cell_specimen_id', 'image_id'])[val].mean()-dfc_noocclusion.groupby(['cell_specimen_id', 'image_id'])[val].mean())/dfc_occlusion.groupby(['cell_specimen_id', 'image_id'])[val].std()
return occlusion_zscore
In [214]:
occlusion_values = occlusion_response_index_bycell(ocdf, val='mean_response')
In [215]:
oc = pd.DataFrame(occlusion_values)
oc = oc.rename(columns={'mean_response':'occlusion_index_zscore'})
In [220]:
metric_df = pd.merge(odf, oc, left_on=['cell_specimen_id', 'image_id'], right_on=['cell_specimen_id', 'image_id'])
In [ ]:
In [222]:
df = metric_df.copy()
df = df[df.pref_stim==True]
df = df[(df.oddball_response_index<2)&(df.occlusion_index_zscore<40)]
sns.scatterplot(data=df, x='occlusion_index_zscore', y='oddball_response_index')
Out[222]:
In [236]:
ocdf_metric = pd.merge(ocdf, oc, left_on=['cell_specimen_id', 'image_id'], right_on=['cell_specimen_id', 'image_id'])
In [237]:
ocdf_metric
Out[237]:
In [240]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
metric = 'occlusion_index_zscore'
df = ocdf_metric.copy()
# df = df[df.oddball==True]
df = df[(df[metric]>-5)&(df[metric]<10)]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
df = df[df.fraction_significant_trials>0.1]
colors = [sns.color_palette()[0], sns.color_palette()[3], sns.color_palette()[2]]
fig, ax = plt.subplots(1, figsize=(5,5), sharey=True, sharex=True)
for i, area in enumerate(['VISp', 'VISpm', 'RSP']):
ax = sns.distplot(df[(df.area==area)&(df.retrogradely_labeled==label)][metric].values, bins=30,
color=colors[i], ax=ax, label=area)
ax.set_xlabel(metric)
legend = ax.legend(fontsize='x-small', title='area', loc='upper right')
plt.setp(legend.get_title(),fontsize='x-small')
ax.set_ylabel('density')
# ax.set_title(area)
In [241]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
metric = 'occlusion_index_zscore'
df = ocdf_metric.copy()
# df = df[df.oddball==True]
df = df[(df[metric]>-5)&(df[metric]<10)]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
df = df[df.fraction_significant_trials>0.1]
colors = [sns.color_palette()[0], sns.color_palette()[3]]
fig, ax = plt.subplots(1,3, figsize=(15,5), sharey=True, sharex=True)
for i, area in enumerate(['VISp', 'VISpm', 'RSP']):
for c, label in enumerate(df.retrogradely_labeled.unique()):
ax[i] = sns.distplot(df[(df.area==area)&(df.retrogradely_labeled==label)][metric].values, bins=30,
color=colors[c], ax=ax[i], label=label)
ax[i].set_xlabel(metric)
legend = ax[i].legend(fontsize='x-small', title='retrogradely labeled', loc='upper right')
plt.setp(legend.get_title(),fontsize='x-small')
ax[i].set_title(area)
ax[0].set_ylabel('density')
Out[241]:
In [242]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
metric = 'oddball_response_index'
metric = 'occlusion_index_zscore'
df = ocdf_metric.copy()
# df = df[df.oddball==True]
df = df[(df[metric]>-5)&(df[metric]<10)]
df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
df = df[df.fraction_significant_trials>0.1]
fig, ax = plt.subplots(1,3, figsize=(15,5), sharey=True, sharex=True)
for i, area in enumerate(['VISp', 'VISpm', 'RSP']):
for c, depth in enumerate(df.depth.unique()):
ax[i] = sns.distplot(df[(df.area==area)&(df.depth==depth)][metric].values, bins=30,
color=colors[c], ax=ax[i], label=depth)
ax[i].set_xlabel(metric)
ax[i].legend()
ax[i].set_title(area)
ax[0].set_ylabel('density')
Out[242]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [231]:
def plot_occlusion_responses(ocdf, metric_df, cell_specimen_id, save_dir=None):
images = oddball_images
colors = sns.color_palette('Blues')
figsize=(20,2.5)
fig, ax = plt.subplots(1, len(images), figsize=figsize, sharey=True)
for i,image_id in enumerate(images):
cell_ocdf = ocdf[(ocdf.cell_specimen_id==cell_specimen_id)&(ocdf.image_id==image_id)]
for o, fraction_occlusion in enumerate(np.sort(ocdf.fraction_occlusion.unique())):
occ_df = cell_ocdf[cell_ocdf.fraction_occlusion==fraction_occlusion]
ax[i].plot(occ_df.mean_trace.values[0], color=colors[o], label=fraction_occlusion)
xticks, xticklabels = sf.get_xticks_xticklabels(cell_ocdf.mean_trace.values[0], 31.)
ax[i].set_xticks(xticks)
ax[i].set_xticklabels(xticklabels)
ax[i].set_xlabel('time (s)')
metric_value = metric_df[(metric_df.cell_specimen_id==cell_specimen_id)&(metric_df.image_id==image_id)].occlusion_index_zscore.values[0]
ax[i].set_title(str(np.round(metric_value,2))+', '+str(np.round(cell_ocdf.fraction_significant_trials.values[0], 2)),fontsize=14)
ax[0].set_ylabel('dF/F')
legend = ax[i].legend(fontsize='x-small', title='fraction occlusion', bbox_to_anchor=(1,1))
plt.setp(legend.get_title(),fontsize='x-small')
fig.tight_layout()
fig.subplots_adjust(bottom=0.25)
fig.subplots_adjust(right=0.8)
if save_dir:
sf.save_figure(fig, figsize, save_dir, 'occlusion_responses', str(int(cell_specimen_id)))
plt.close()
In [232]:
plot_occlusion_responses(ocdf, metric_df, cell_specimen_id, save_dir=None)
In [235]:
for cell_specimen_id in ocdf.cell_specimen_id.unique():
plot_occlusion_responses(ocdf, metric_df, cell_specimen_id, save_dir=save_dir)
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [116]:
odf = odf.reset_index()
occlusion_index = []
for index in odf.index.values:
row_df = odf.iloc[index]
cell_specimen_id = row_df.cell_specimen_id
image_id = row_df.image_id
occlusion_mean_response = row_df.mean_response
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)&(tdf.second_in_sequence==True)]
transitions_mean_response = tcdf.mean_response.values[0]
occlusion_response_index = ((oddball_mean_response-transitions_mean_response)/(oddball_mean_response+transitions_mean_response))
occlusion_index.append(occlusion_response_index)
odf['occlusion_response_index'] = occlusion_index
Out[116]:
In [ ]:
In [ ]:
ocdf = ocdf.reset_index()
oddball_index = []
for index in ocdf.index.values:
row_df = ocdf.iloc[index]
cell_specimen_id = row_df.cell_specimen_id
image_id = row_df.image_id
oddball_mean_response = row_df.mean_response
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)&(tdf.second_in_sequence==True)]
transitions_mean_response = tcdf.mean_response.values[0]
oddball_response_index = ((oddball_mean_response-transitions_mean_response)/(oddball_mean_response+transitions_mean_response))
oddball_index.append(oddball_response_index)
odf['occlusion_response_index'] = oddball_index
In [ ]:
In [ ]:
In [ ]:
df = ocdf.copy()
df_shallowVISp = df[(df['targeted_structure']=='VISp')&(df['imaging_depth']<=300)]
df_focus = df_shallowVISp
del df_shallowVISp
responsive_cells = df_focus.cell_specimen_id.unique()
for ii in range(0,len(responsive_cells)):
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]])<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if not 0 in list(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion']):
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion'].unique())<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
ori_bycell=occlusion_response_index_bycell(df_focus)
ori_shallowVISp=ori_bycell.dropna()
# In[]:
df_deepVISp = df[(df['targeted_structure']=='VISp')&(df['imaging_depth']>300)]
df_focus = df_deepVISp
del df_deepVISp
responsive_cells = df_focus.cell_specimen_id.unique()
for ii in range(0,len(responsive_cells)):
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]])<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if not 0 in list(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion']):
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion'].unique())<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
ori_bycell=occlusion_response_index_bycell(df_focus)
ori_deepVISp=ori_bycell.dropna()
# In[]:
df_shallowVISpm = df[(df['targeted_structure']=='VISpm')&(df['imaging_depth']<=300)]
df_focus = df_shallowVISpm
del df_shallowVISpm
responsive_cells = df_focus.cell_specimen_id.unique()
for ii in range(0,len(responsive_cells)):
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]])<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if not 0 in list(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion']):
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion'].unique())<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
ori_bycell=occlusion_response_index_bycell(df_focus)
ori_shallowVISpm=ori_bycell.dropna()
# In[]:
df_deepVISpm = df[(df['targeted_structure']=='VISpm')&(df['imaging_depth']>300)]
df_focus = df_deepVISpm
del df_deepVISpm
responsive_cells = df_focus.cell_specimen_id.unique()
for ii in range(0,len(responsive_cells)):
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]])<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if not 0 in list(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion']):
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion'].unique())<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
ori_bycell=occlusion_response_index_bycell(df_focus)
ori_deepVISpm=ori_bycell.dropna()
# In[]:
#cache_dir = r'\\allen\programs\braintv\workgroups\nc-ophys\opc\opc_analysis'
#cache_dir = r"C:\Users\hannahc\Desktop\opc_data"
df_shallowRSP = df[(df['targeted_structure']=='RSP')&(df['imaging_depth']<=300)]
df_focus = df_shallowRSP
del df_shallowRSP
responsive_cells = df_focus.cell_specimen_id.unique()
for ii in range(0,len(responsive_cells)):
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]])<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if not 0 in list(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion']):
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion'].unique())<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
ori_bycell=occlusion_response_index_bycell(df_focus)
ori_shallowRSP=ori_bycell.dropna()
# In[]:
df_deepRSP = df[(df['targeted_structure']=='VISp')&(df['imaging_depth']>300)]
df_focus = df_deepRSP
del df_deepRSP
responsive_cells = df_focus.cell_specimen_id.unique()
for ii in range(0,len(responsive_cells)):
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]])<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if not 0 in list(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion']):
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
if len(df_focus[df_focus['cell_specimen_id']==responsive_cells[ii]]['fraction_occlusion'].unique())<3:
df_focus = df_focus[df_focus['cell_specimen_id']!=responsive_cells[ii]]
ori_bycell=occlusion_response_index_bycell(df_focus)
ori_deepRSP=ori_bycell.dropna()
# In[]:
areas = [['shallowVISp']*len(ori_shallowVISp)+ ['deepVISp']*len(ori_deepVISp)
+['shallowVISpm']*len(ori_shallowVISpm)+ ['deepVISpm']*len(ori_deepVISpm)+['shallowRSP']*len(ori_shallowRSP)+ ['deepRSP']*len(ori_deepRSP)]
areas =np.array(areas[0])
all_ori = ori_shallowVISp.append(ori_deepVISp)
all_ori= all_ori.append(ori_shallowVISpm)
all_ori= all_ori.append(ori_deepVISpm)
all_ori= all_ori.append(ori_shallowRSP)
all_ori= all_ori.append(ori_deepRSP)
df_ori = pd.DataFrame({'ORI':all_ori,'area':areas} )
# In[]:
ori_thresh = 1
df_occlusion_pref = df_ori[df_ori.ORI>=1]
n_occlusion_pref = df_occlusion_pref.groupby('area').count()
df_occlusion_nonpref = df_ori[df_ori.ORI<1]
n_occlusion_nonpref = df_occlusion_nonpref.groupby('area').count()
n_occlusion_pref = n_occlusion_pref.rename(columns={'ORI': 'occlusion preferred'})
n_occlusion_nonpref = n_occlusion_nonpref.rename(columns={'ORI': 'occlusion not preferred'})
newdf =pd.concat([n_occlusion_pref,n_occlusion_nonpref], axis=1)
In [ ]:
# In[]:
arealist = list(newdf.index.unique())
fig,ax = plt.subplots(figsize=(9,4))
ax.boxplot([df_ori.ORI[df_ori.area==arealist[0]].values,df_ori.ORI[df_ori.area==arealist[1]].values,
df_ori.ORI[df_ori.area==arealist[2]].values,df_ori.ORI[df_ori.area==arealist[3]].values,
df_ori.ORI[df_ori.area==arealist[4]].values,df_ori.ORI[df_ori.area==arealist[5]].values])
ax.set_xticks(np.arange(1,7,1))
ax.set_xticklabels(tuple(arealist),fontsize = 12)
ax.set_ylim([0,50])
ax.set_ylabel('Occlusion Response Index',fontsize = 14)
plt.show()
# In[]:
import seaborn as sns
%matplotlib inline
#ax = sns.scatterplot(x="areas", y="ORI", hue="areas",data=df_ori)
fig,ax = plt.subplots()
arealist = list(newdf.index.unique())
colors = ["red"]
jj = 0
for i in arange(1,7,1):
y = df_ori.ORI[df_ori.area==arealist[jj]]
# Add some random "jitter" to the x-axis
x = np.random.normal(i, 0.03, size=len(y))
ax.scatter(x, y, c=colors[0], alpha=0.5)
jj = jj+1
ax.boxplot([df_ori.ORI[df_ori.area==arealist[0]].values,df_ori.ORI[df_ori.area==arealist[1]].values,
df_ori.ORI[df_ori.area==arealist[2]].values,df_ori.ORI[df_ori.area==arealist[3]].values,
df_ori.ORI[df_ori.area==arealist[4]].values,df_ori.ORI[df_ori.area==arealist[5]].values])
ax.set_xticks(arealist)
ax.set_ylabel('Occlusion Response Index')
plt.show()
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
pick a cell and plot stuff¶
In [89]:
save_dir = os.path.join(cache_dir, 'multi_session_summary_figures')
In [25]:
cell_specimen_id = cell_specimen_ids[10]
In [86]:
cell_specimen_id = responsive_cells[1]
In [90]:
figsize = (5,4)
fig, ax = plt.subplots(figsize=figsize)
for image_id in images[::-1]:
cdf = odf[(odf.image_id==image_id)&(odf.cell_specimen_id==cell_specimen_id)]
mean_trace = cdf.mean_trace.values[0]
if image_id in sequence_images:
ind = np.where(sequence_images==image_id)[0][0]
color = sequence_colors[ind]
else:
ind = np.where(oddball_images==image_id)[0][0]
color = oddball_colors[ind]
ax.plot(mean_trace, color=color)
xticks, xticklabels = sf.get_xticks_xticklabels(mean_trace, 31., interval_sec=1)
ax.set_xticks(xticks)
ax.set_xticklabels([int(x) for x in xticklabels])
ax.set_ylabel('dF/F')
ax.set_xlabel('time (sec)')
ax.set_title(cell_specimen_id)
In [93]:
for cell_specimen_id in cell_specimen_ids:
figsize = (5,4)
fig, ax = plt.subplots(figsize=figsize)
for image_id in images[::-1]:
cdf = odf[(odf.image_id==image_id)&(odf.cell_specimen_id==cell_specimen_id)]
mean_trace = cdf.mean_trace.values[0]
if image_id in sequence_images:
ind = np.where(sequence_images==image_id)[0][0]
color = sequence_colors[ind]
else:
ind = np.where(oddball_images==image_id)[0][0]
color = oddball_colors[ind]
ax.plot(mean_trace, color=color)
xticks, xticklabels = sf.get_xticks_xticklabels(mean_trace, 31., interval_sec=1)
ax.set_xticks(xticks)
ax.set_xticklabels([int(x) for x in xticklabels])
ax.set_ylabel('dF/F')
ax.set_xlabel('time (sec)')
ax.set_title(cell_specimen_id)
sf.save_figure(fig, figsize, save_dir, 'oddball_mean_trace', str(cell_specimen_id))
plt.close()
In [94]:
figsize = (14,5)
fig, ax = plt.subplots(2,7,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
for i,image_id in enumerate(images):
cdf = odf[(odf.image_id==image_id)&(odf.cell_specimen_id==cell_specimen_id)]
mean_trace = cdf.mean_trace.values[0]
ax[i].plot(mean_trace, color='green', label='oddball block')
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)]
mean_trace = tcdf.mean_trace.values[0]
ax[i].plot(mean_trace, color='magenta', label='transition control block')
xticks, xticklabels = sf.get_xticks_xticklabels(mean_trace, 31., interval_sec=1)
ax[i].set_xticks(xticks)
ax[i].set_xticklabels(xticklabels)
ax[i].set_title(str(int(image_id)))
ax[0].set_ylabel('dF/F')
ax[7].set_ylabel('dF/F')
for i in range(7,14):
ax[i].set_xlabel('time (sec)')
fig.tight_layout()
plt.suptitle(str(cell_specimen_id), x=0.525, y=.97, fontsize=18)
plt.subplots_adjust(top=0.85)
In [95]:
for cell_specimen_id in cell_specimen_ids:
figsize = (14,5)
fig, ax = plt.subplots(2,7,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
for i,image_id in enumerate(images):
cdf = odf[(odf.image_id==image_id)&(odf.cell_specimen_id==cell_specimen_id)]
mean_trace = cdf.mean_trace.values[0]
ax[i].plot(mean_trace, color='green', label='oddball block')
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)]
mean_trace = tcdf.mean_trace.values[0]
ax[i].plot(mean_trace, color='magenta', label='transition control block')
xticks, xticklabels = sf.get_xticks_xticklabels(mean_trace, 31., interval_sec=1)
ax[i].set_xticks(xticks)
ax[i].set_xticklabels(xticklabels)
ax[i].set_title(str(int(image_id)))
ax[0].set_ylabel('dF/F')
ax[7].set_ylabel('dF/F')
for i in range(7,14):
ax[i].set_xlabel('time (sec)')
fig.tight_layout()
plt.suptitle(str(cell_specimen_id), x=0.525, y=.97, fontsize=18)
plt.subplots_adjust(top=0.85)
sf.save_figure(fig, figsize, save_dir, 'oddball_transition_mean_trace', str(cell_specimen_id))
plt.close()
In [ ]:
In [76]:
figsize = (6,4)
fig, ax = plt.subplots(figsize=figsize)
for i,image_id in enumerate(images):
cdf = odf[(odf.image_id==image_id)&(odf.cell_specimen_id==cell_specimen_id)]
mean_response = cdf.mean_response.values[0]
ax.plot(i, mean_response, 'o', color='green')
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)]
mean_response = tcdf.mean_response.values[0]
ax.plot(i, mean_response, 'o' ,color='magenta')
cdf = odf[(odf.image_id==image_id)&(odf.cell_specimen_id==cell_specimen_id)]
mean_response = cdf.mean_response.values[0]
ax.plot(i, mean_response, 'o', color='green', label='sequence')
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)]
mean_response = tcdf.mean_response.values[0]
ax.plot(i, mean_response, 'o' ,color='magenta', label='transition control')
ax.set_title(str(int(cell_specimen_id)))
ax.set_ylabel('dF/F')
ax.set_xticks(np.arange(0,len(images)))
ax.set_xticklabels([int(image) for image in images], rotation=90)
ax.set_xlabel('image ID')
ax.legend(fontsize='x-small')
fig.tight_layout()
In [403]:
for cell_specimen_id in cell_specimen_ids:
figsize = (6,4)
fig, ax = plt.subplots(figsize=figsize)
for i,image_id in enumerate(images):
cdf = odf[(odf.image_id==image_id)&(odf.cell_specimen_id==cell_specimen_id)]
mean_response = cdf.mean_response.values[0]
ax.plot(i, mean_response, 'o', color='green')
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)]
mean_response = tcdf.mean_response.values[0]
ax.plot(i, mean_response, 'o' ,color='magenta')
cdf = odf[(odf.image_id==image_id)&(odf.cell_specimen_id==cell_specimen_id)]
mean_response = cdf.mean_response.values[0]
ax.plot(i, mean_response, 'o', color='green', label='sequence')
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)]
mean_response = tcdf.mean_response.values[0]
ax.plot(i, mean_response, 'o' ,color='magenta', label='transition control')
ax.set_title(str(int(cell_specimen_id)))
ax.set_ylabel('dF/F')
ax.set_xticks(np.arange(0,len(images)))
ax.set_xticklabels([int(image) for image in images], rotation=90)
ax.set_xlabel('image ID')
ax.legend(fontsize='x-small')
fig.tight_layout()
sf.save_figure(fig, figsize, save_dir, 'oddball_transition_mean_response', str(cell_specimen_id))
plt.close()
In [ ]:
oddball response metric¶
In [97]:
analysis.get_response_df_dict()
o = analysis.response_df_dict['oddball']
In [101]:
t = analysis.response_df_dict['transition_control']
In [100]:
tmp = o[o.cell_index==1]
for image in images:
print(image, len(tmp[tmp.image_id==image]))
In [102]:
tmp = t[t.cell_index==1]
for image in images:
print(image, len(tmp[tmp.image_id==image]))
In [182]:
odf = oddball_df.copy()
tdf = transitions_df.copy()
In [183]:
odf = odf.reset_index()
odf = odf.drop(columns=['index'])
In [186]:
odf = odf.reset_index()
oddball_index = []
for index in odf.index.values:
row_df = odf.iloc[index]
cell_specimen_id = row_df.cell_specimen_id
image_id = row_df.image_id
oddball_mean_response = row_df.mean_response
tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)]
transitions_mean_response = tcdf.mean_response.values[0]
oddball_response_index = ((oddball_mean_response-transitions_mean_response)/(oddball_mean_response+transitions_mean_response))
oddball_index.append(oddball_response_index)
odf['oddball_response_index'] = oddball_index
odf['area'] = odf.targeted_structure.values
odf['depth'] = ['deep' if depth > 250 else 'superficial' for depth in odf.imaging_depth.values]
odf['location'] = None
for row in odf.index:
odf.loc[row, 'location'] = odf.iloc[row].area+'_'+odf.iloc[row].depth
tdf['area'] = tdf.targeted_structure.values
tdf['depth'] = ['deep' if depth > 250 else 'superficial' for depth in tdf.imaging_depth.values]
tdf['location'] = None
for row in tdf.index:
tdf.loc[row, 'location'] = tdf.iloc[row].area+'_'+tdf.iloc[row].depth
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = odf.copy()
df = df[df.oddball==True]
# df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='oddball_response_index', order=locations, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
In [189]:
In [370]:
In [192]:
In [ ]:
In [380]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = odf.copy()
df = df[df.oddball==True]
# df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='oddball_response_index', order=locations, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
Out[380]:
In [399]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
metric = 'oddball_response_index'
df = odf.copy()
df = df[df.oddball==True]
df = df[(df[metric]>-2)&(df[metric]<2)]
# df = df[df.pref_stim==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots()
for c, area in enumerate(['VISp', 'VISpm', 'RSP']):
ax = sns.distplot(df[df.area==area][metric].values, bins=30,
color=sns.color_palette()[c], ax=ax, label=area)
ax.legend()
ax.set_xlabel(metric)
Out[399]:
In [222]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = odf[(odf.oddball_response_index<2)&(odf.oddball_response_index>-2)].copy()
df = df[df.oddball==True]
# df = df[df.fraction_significant_trials>0.2]
df = df[df.mean_response>0.05]
fig, ax = plt.subplots()
ax = sns.swarmplot(data=df, x='location', y='oddball_response_index', order=locations, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
Out[222]:
habituated vs novel¶
In [ ]:
# # figsize = (6,4)
# # fig, ax = plt.subplots(figsize=figsize)
# oddball_index = []
# for index in odf.index.values:
# row_df = odf.iloc[index]
# cell_specimen_id = row_df.cell_specimen_id
# image_id = row_df.image_id
# oddball_mean_response = row_df.mean_response
# tcdf = tdf[(tdf.image_id==image_id)&(tdf.cell_specimen_id==cell_specimen_id)]
# habituated_mean_response = tcdf.mean_response.mean()
# oddball_response_index = ((oddball_mean_response-transitions_mean_response)/(oddball_mean_response+transitions_mean_response))
# oddball_index.append(oddball_response_index)
# odf['oddball_response_index'] = oddball_index
In [ ]:
In [ ]:
mean response across experiments¶
In [ ]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
In [104]:
experiment_ids = odf.experiment_id.unique()
experiment_id = experiment_ids[4]
In [338]:
len(experiment_ids)/8
Out[338]:
In [ ]:
In [341]:
figsize=(20,15)
fig, ax = plt.subplots(5,7,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
i=0
for area in odf.targeted_structure.unique()[::-1]:
tmp = odf[odf.targeted_structure==area]
for depth in np.sort(tmp.imaging_depth.unique()):
tmp2 = tmp[tmp.imaging_depth==depth]
for experiment_id in tmp2.experiment_id.unique():
edf = odf[odf.experiment_id==experiment_id]
sequence = edf[edf.image_id==images[3]].mean_trace.mean()
oddballs = edf[edf.oddball==True].mean_trace.mean()
ax[i].plot(sequence, color='blue', label='sequence')
ax[i].plot(oddballs, color='red', label='oddballs')
xticks, xticklabels = sf.get_xticks_xticklabels(sequence, 31., interval_sec=1)
ax[i].set_xticks(xticks)
ax[i].set_xticklabels([int(x) for x in xticklabels])
ax[i].set_ylabel('dF/F')
ax[i].set_xlabel('time (sec)')
ax[i].set_title(edf.targeted_structure.values[0]+', '+str(edf.imaging_depth.values[0]))
i+=1
fig.tight_layout()
In [401]:
figsize=(20,15)
fig, ax = plt.subplots(5,7,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
i=0
for area in odf.targeted_structure.unique()[::-1]:
tmp = odf[odf.targeted_structure==area]
for depth in np.sort(tmp.imaging_depth.unique()):
tmp2 = tmp[tmp.imaging_depth==depth]
for experiment_id in tmp2.experiment_id.unique():
dataset = OpenScopePredictiveCodingDataset(experiment_id, cache_dir)
ax[i].imshow(dataset.max_projection, cmap='gray', vmax=np.amax(dataset.max_projection)/2)
ax[i].axis('off')
ax[i].set_title(dataset.metadata.targeted_structure.values[0]+', '+str(dataset.metadata.imaging_depth.values[0]))
i+=1
fig.tight_layout()
In [365]:
figsize=(25,15)
fig, ax = plt.subplots(5,7,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
i=0
for area in odf.targeted_structure.unique()[::-1]:
tmp = odf[odf.targeted_structure==area]
for depth in np.sort(tmp.imaging_depth.unique()):
tmp2 = tmp[tmp.imaging_depth==depth]
for experiment_id in tmp2.experiment_id.unique():
dataset = OpenScopePredictiveCodingDataset(experiment_id, cache_dir)
ax[i] = esf.plot_traces_heatmap(dataset.dff_traces, ax=ax[i], save_dir=None)
ax[i].axis('off')
ax[i].set_title(tmp2.targeted_structure.values[0]+', '+str(tmp2.imaging_depth.values[0]))
i+=1
fig.tight_layout()
In [352]:
def plot_mean_image_response_heatmap(mean_df, image_ids, colorbar=True, title=None, ax=None, save_dir=None):
df = mean_df.copy()
images = np.sort(df.image_id.unique())
images = image_ids
cell_list = []
for image in images:
tmp = df[(df.image_id == image) & (df.pref_stim == True)]
order = np.argsort(tmp.mean_response.values)[::-1]
cell_ids = list(tmp.cell_specimen_id.values[order])
cell_list = cell_list + cell_ids
response_matrix = np.empty((len(cell_list), len(images)))
for i, cell_specimen_id in enumerate(cell_list):
responses = []
for image in images:
response = df[(df.cell_specimen_id == cell_specimen_id) & (df.image_id == image)].mean_response.values[0]
responses.append(response)
response_matrix[i, :] = np.asarray(responses)
if ax is None:
figsize = (5, 8)
fig, ax = plt.subplots(figsize=figsize)
ax = sns.heatmap(response_matrix, cmap='magma', linewidths=0, linecolor='white', square=False,
vmin=0, vmax=0.3, robust=True, cbar=colorbar,
cbar_kws={"drawedges": False, "shrink": 1, "label": "mean dF/F"}, ax=ax)
if title is None:
title = 'mean response by image'
ax.set_title(title, va='bottom', ha='center')
ax.set_xticks(np.arange(0,len(images),1))
ax.set_xticklabels([int(image) for image in images], rotation=90)
ax.set_ylabel('cells')
interval = 10
ax.set_yticks(np.arange(0, response_matrix.shape[0], interval))
ax.set_yticklabels(np.arange(0, response_matrix.shape[0], interval))
if save_dir:
fig.tight_layout()
sf.save_figure(fig, figsize, save_dir, 'mean_image_response_heatmap', str(experiment_id)+'_'+title)
return ax
In [123]:
image_ids = analysis.get_image_ids()
for experiment_id in experiment_ids:
edf = odf[odf.experiment_id==experiment_id]
title = edf.targeted_structure.values[0]+', '+str(edf.imaging_depth.values[0])
plot_mean_image_response_heatmap(edf, image_ids, title=title, ax=None, save_dir=save_dir)
In [353]:
image_ids = analysis.get_image_ids()
figsize=(20,15)
fig, ax = plt.subplots(5,7,figsize=figsize, sharex=False, sharey=False)
ax = ax.ravel()
i=0
for area in odf.targeted_structure.unique()[::-1]:
tmp = odf[odf.targeted_structure==area]
for depth in np.sort(tmp.imaging_depth.unique()):
tmp2 = tmp[tmp.imaging_depth==depth]
for experiment_id in tmp2.experiment_id.unique():
edf = odf[odf.experiment_id==experiment_id]
title = edf.targeted_structure.values[0]+', '+str(edf.imaging_depth.values[0])
ax[i] = plot_mean_image_response_heatmap(edf, image_ids, colorbar=False, title=title, ax=ax[i], save_dir=None)
ax[i].set_xlabel('')
ax[i].set_xticklabels('')
ax[i].set_yticklabels('')
ax[i].set_ylabel('')
i+=1
fig.tight_layout()
In [402]:
image_ids = analysis.get_image_ids()
figsize=(20,15)
fig, ax = plt.subplots(7,8,figsize=figsize, sharex=False, sharey=False)
ax = ax.ravel()
i=0
for area in tdf.targeted_structure.unique()[::-1]:
tmp = tdf[tdf.targeted_structure==area]
for depth in np.sort(tmp.imaging_depth.unique()):
tmp2 = tmp[tmp.imaging_depth==depth]
for experiment_id in tmp2.experiment_id.unique():
edf = tdf[tdf.experiment_id==experiment_id]
title = edf.targeted_structure.values[0]+', '+str(edf.imaging_depth.values[0])
ax[i] = plot_mean_image_response_heatmap(edf, image_ids, colorbar=False, title=title, ax=ax[i], save_dir=None)
ax[i].set_xlabel('')
ax[i].set_xticklabels('')
ax[i].set_yticklabels('')
ax[i].set_ylabel('')
i+=1
fig.tight_layout()
In [360]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = odf.copy()
df = df[df.oddball==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='mean_response', order=locations, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
ax.set_title('response magnitude during sequence block')
Out[360]:
In [366]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = tdf.copy()
df = df[df.oddball==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='mean_response', order=locations, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
ax.set_title('response magnitude during transition control')
Out[366]:
In [367]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = odf.copy()
df = df[df.oddball==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='fraction_significant_trials', order=locations, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
ax.set_title('reliability during sequence block')
Out[367]:
In [369]:
locations = ['VISp_superficial', 'VISp_deep', 'VISpm_superficial', 'VISpm_deep', 'RSP_superficial', 'RSP_deep']
df = tdf.copy()
df = df[df.oddball==True]
# df = df[df.mean_response>0.05]
# df = df[df.fraction_significant_trials>0.0]
fig, ax = plt.subplots()
ax = sns.pointplot(data=df, x='location', y='fraction_significant_trials', order=locations, ax=ax)
ax.set_xticklabels(locations, rotation=90)
ax.set_xlabel('')
ax.set_title('reliability during transition control')
Out[369]:
image responsiveness¶
In [293]:
def get_session_summary_df(fdf, use_events=False):
# only consider responses to preferred stimulus
df = fdf[fdf.pref_stim==True].copy()
summary_data = []
for experiment_id in df.experiment_id.unique():
expt = df[(df.experiment_id == experiment_id)]
area = expt.targeted_structure.unique()[0]
if 'location' in fdf.keys():
location = expt.location.unique()[0]
else:
lcation = None
cre_line = expt.cre_line.unique()[0]
mouse_id = expt.donor_id.unique()[0]
depth = expt.imaging_depth.unique()[0]
session_type = expt.session_type.unique()[0]
image_set = expt.image_set.unique()[0]
experiment_date = expt.experiment_date.unique()[0]
n_cells = len(expt.cell_specimen_id.unique())
summary_data.append(
[experiment_id, mouse_id, session_type, image_set, area, location, cre_line, experiment_date, depth, n_cells])
columns = ['experiment_id', 'mouse_id', 'session_type','image_set', 'area', 'location', 'cre_line',
'experiment_date', 'depth', 'n_cells']
session_summary_df = pd.DataFrame(data=summary_data, columns=columns)
return session_summary_df
def add_metric_to_session_summary_df(metric, data, session_summary_df):
df = data.copy()
session_summary_df[metric] = None
for experiment_id in session_summary_df.experiment_id.unique():
metric_values = df[(df.experiment_id==experiment_id)][metric].values
session_mean = np.nanmean(metric_values)
idx = session_summary_df[(session_summary_df.experiment_id==experiment_id)].index[0]
session_summary_df.loc[idx, metric] = session_mean
return session_summary_df
In [294]:
odf['image_set'] = 'A'
session_summary_df = get_session_summary_df(odf)
session_summary_df
Out[294]:
In [295]:
df = session_summary_df.copy()
figsize = (5,4)
fig, ax = plt.subplots(figsize=figsize)
ax = sns.stripplot(data=df, x='area', y='n_cells', hue='image_set', ax = ax, dodge=True)
ax.set_ylabel('# cells per session')
fig.tight_layout()
sf.save_figure(fig, figsize, save_dir, 'session_numbers', 'n_cells_by_image_set')
In [296]:
def adjust_box_widths(ax, fac):
from matplotlib.patches import PathPatch
# Adjust the withs of a seaborn-generated boxplot.
for c in ax.get_children():
if isinstance(c, PathPatch):
p = c.get_path()
verts = p.vertices
verts_sub = verts[:-1]
xmin = np.min(verts_sub[:, 0])
xmax = np.max(verts_sub[:, 0])
xmid = 0.5 * (xmin + xmax)
xhalf = 0.5 * (xmax - xmin)
xmin_new = xmid - fac * xhalf
xmax_new = xmid + fac * xhalf
verts_sub[verts_sub[:, 0] == xmin, 0] = xmin_new
verts_sub[verts_sub[:, 0] == xmax, 0] = xmax_new
for l in ax.lines:
if np.all(l.get_xdata() == [xmin, xmax]):
l.set_xdata([xmin_new, xmax_new])
def plot_boxplot_for_condition(df, metric, condition='image_set', condition_values=['A', 'B', 'C', 'D'],
colors=sns.color_palette(), hue='cre_line', ylabel=None,
range=(0, 1), ax=None, save_figures=False, save_dir=None, folder=None):
if ax is None:
figsize = (4.5, 4.5)
fig, ax = plt.subplots(figsize=figsize)
ax = sns.boxplot(data=df, x=condition, y=metric,
hue=hue, ax=ax, width=0.4, dodge=True, palette=colors)
if ylabel is None:
ax.set_ylabel('fraction of cells per session')
else:
ax.set_ylabel(ylabel)
ax.set_ylim(range[0] - 0.05, range[1] + .05)
ax.get_legend().remove()
ax.set_title(metric)
sns.despine(offset=10, trim=True)
plt.gcf().subplots_adjust(top=0.85)
plt.gcf().subplots_adjust(left=0.25)
plt.gcf().subplots_adjust(right=0.85)
plt.gcf().subplots_adjust(bottom=0.25)
if save_figures:
psf.save_figure(fig, figsize, save_dir, folder, metric + '_by_' + condition + '_box')
return ax
In [297]:
def plot_boxplot_and_swarm_for_condition(df, metric, condition='cre_line', condition_values=None,
colors=sns.color_palette(), hue='image_set', hue_order=None, order=None, ylabel=None, xlabel=None, title=None,
plot_swarm=True, show_stats=True, range=(0,1), ax=None, save_figures=False, save_dir=None, folder=None, suffix=''):
df[metric] = pd.to_numeric(df[metric])
if hue_order is None:
hue_order = np.sort(df[hue].unique())
if ax is None:
figsize = (4.5,4.5)
fig,ax = plt.subplots(figsize=figsize)
if order is None:
order = condition_values
ax = sns.boxplot(data=df, x=condition, y=metric, hue_order=hue_order, order=order,
hue=hue, ax=ax, dodge=True, palette=colors) #color='white',
adjust_box_widths(ax, 0.8)
if plot_swarm:
ax = sns.swarmplot(data=df, x=condition, y=metric, order=order,
size=3, ax=ax, hue=hue, hue_order=hue_order, color='.3', dodge=True) #palette=colors,
swarm_cols = ax.collections
for swarm in swarm_cols:
swarm.set_facecolors([0.6,0.6,0.6])
if ylabel is None:
ax.set_ylabel('fraction of cells per session')
else:
ax.set_ylabel(ylabel)
if xlabel is None:
if len(condition.split('_'))>1:
ax.set_xlabel(condition.split('_')[0]+' '+condition.split('_')[1])
else:
ax.set_xlabel(condition)
else:
ax.set_xlabel(xlabel)
ax.set_xticklabels(condition_values, rotation=90)
if title is None:
ax.set_title(metric)
else:
ax.set_title(title)
ax.set_ylim(range[0]-0.05,range[1]+.05)
ax.get_legend().remove()
sns.despine(offset=10, trim=True)
if show_stats:
hue_names = hue_order
sig_y_val = df[metric].max()
stats_df = get_stats_for_conditions(df, metric, condition, condition_values, hue, hue_names)
label_df = get_label_df(ax, stats_df, condition, condition_values, hue, hue_names)
ax = label_ax_with_stats(ax, label_df, stats_df, sig_y_val, condition, condition_values, hue, hue_names)
ax.set_ylim(range[0]-0.05,range[1]+.05)
plt.gcf().subplots_adjust(top=0.85)
plt.gcf().subplots_adjust(left=0.25)
plt.gcf().subplots_adjust(right=0.85)
plt.gcf().subplots_adjust(bottom=0.25)
if save_figures:
# l = ax.legend(title=condition, fontsize='small')
# plt.setp(l.get_title(),fontsize='small')
sf.save_figure(fig ,figsize, save_dir, folder, metric+'_by_'+condition+'_box_swarm'+suffix)
return ax
In [298]:
df = odf.copy()
df = df[df.pref_stim==True]
metric = 'fraction_significant_trials'
fraction_responsive_list = []
for experiment_id in session_summary_df.experiment_id.unique():
expt = df[(df.experiment_id == experiment_id)]
n_cells = len(expt.cell_specimen_id.unique())
if np.isnan(expt.fraction_significant_trials.values[0])==False: #if has omitted
n_responsive = len(expt[expt.fraction_significant_trials>0.25])
else:
n_responsive = np.nan
fraction = n_responsive / float(n_cells)
fraction_responsive_list.append(fraction)
session_summary_df['fraction_responsive_'+metric] = fraction_responsive_list
In [405]:
session_summary_df['fraction_responsive_oddball'] = session_summary_df.fraction_responsive_fraction_significant_trials.values
In [406]:
areas = ['VISp', 'VISpm', 'RSP']
In [408]:
metric = 'fraction_responsive_oddball'
xlabel = ''
title = 'fraction responsive cells\nsequence block'
ylabel = 'fraction responsive cells'
hue = 'area'
hue_order = np.sort(session_summary_df.area.unique())
colors = sns.color_palette()
df = session_summary_df.copy()
df[metric] = pd.to_numeric(df[metric])
# df['cre line'] = [cre_line.split('-')[0] for cre_line in df.cre_line.values]
condition = 'area'
condition_values = np.sort(df['area'].unique())
plot_boxplot_and_swarm_for_condition(df, metric, condition, condition_values, colors, hue, hue_order=hue_order,
plot_swarm=True, ylabel=ylabel, xlabel=xlabel, title=title, range=(0,1), ax=None,
order=areas, suffix='', save_figures=True, save_dir=save_dir, folder='fraction_responsive',
show_stats=False)
Out[408]:
In [409]:
metric = 'fraction_responsive'
xlabel = ''
title = 'fraction responsive cells\nsequences'
ylabel = 'fraction responsive cells'
hue = 'area'
hue_order = np.sort(session_summary_df.area.unique())
colors = sns.color_palette()
df = session_summary_df.copy()
df[metric] = pd.to_numeric(df[metric])
# df['cre line'] = [cre_line.split('-')[0] for cre_line in df.cre_line.values]
condition = 'area'
condition_values = np.sort(df['area'].unique())
order = areas
fig, ax = plt.subplots(figsize=(4,4))
ax = sns.boxplot(data=df, x=condition, y=metric, hue_order=hue_order, order=order,
hue=hue, ax=ax, dodge=False, palette=colors, width=0.3) #color='white',
ax = sns.swarmplot(data=df, x=condition, y=metric, order=order,
size=3, ax=ax, hue=hue, hue_order=hue_order, color='.3', dodge=False) #palette=colors,
ax.legend_.remove()
ax.set_ylim(0,1)
ax.set_xlabel('')
ax.set_title(title)
sns.despine()
In [ ]:
In [309]:
metric = 'oddball_response_index'
xlabel = ''
title = 'oddball response index'
ylabel = title
hue = 'location'
hue_order = locations
colors = sns.color_palette()
order = locations
df = odf.copy()
df[metric] = pd.to_numeric(df[metric])
df = df[(df[metric]<2)&(df[metric]>-2)]
df = df[df.mean_response>0.05]
# df['cre line'] = [cre_line.split('-')[0] for cre_line in df.cre_line.values]
condition = 'location'
condition_values = np.sort(df['location'].unique())
fig, ax = plt.subplots()
ax = sns.boxenplot(data=df, y=condition, x=metric, hue_order=hue_order, order=order,
hue=hue, ax=ax, dodge=True, palette=colors, orient='h') #color='white',
ax.legend_.remove()
ax.set_ylabel('')
# plot_boxplot_and_swarm_for_condition(df, metric, condition, condition_values, colors, hue, hue_order=hue_order,
# order=None, plot_swarm=False, ylabel=ylabel, xlabel=xlabel, title=title, range=(-2,2), ax=None,
# suffix='', save_figures=True, save_dir=save_dir, folder='fraction_responsive',
# show_stats=False)
Out[309]:
fraction responsive across stimulus types¶
In [358]:
analysis.response_df_dict.keys()
Out[358]:
In [371]:
df = tdf.copy()
df = df[df.pref_stim==True]
metric = 'fraction_significant_trials'
fraction_responsive_list = []
for experiment_id in session_summary_df.experiment_id.unique():
expt = df[(df.experiment_id == experiment_id)]
n_cells = len(expt.cell_specimen_id.unique())
if np.isnan(expt.fraction_significant_trials.values[0])==False: #if has omitted
n_responsive = len(expt[expt.fraction_significant_trials>0.25])
else:
n_responsive = np.nan
fraction = n_responsive / float(n_cells)
fraction_responsive_list.append(fraction)
session_summary_df['fraction_responsive_'+metric] = fraction_responsive_list
In [375]:
session_summary_df['fraction_responsive_transition'] = session_summary_df.fraction_responsive_fraction_significant_trials.values
In [376]:
metric = 'fraction_responsive_transition'
xlabel = ''
title = 'fraction responsive cells\ntransition control'
ylabel = 'fraction responsive cells'
hue = 'area'
hue_order = np.sort(session_summary_df.area.unique())
colors = sns.color_palette()
df = session_summary_df.copy()
df[metric] = pd.to_numeric(df[metric])
# df['cre line'] = [cre_line.split('-')[0] for cre_line in df.cre_line.values]
condition = 'area'
condition_values = np.sort(df['area'].unique())
order = areas
fig, ax = plt.subplots(figsize=(4,4))
ax = sns.boxplot(data=df, x=condition, y=metric, hue_order=hue_order, order=order,
hue=hue, ax=ax, dodge=False, palette=colors, width=0.3) #color='white',
ax = sns.swarmplot(data=df, x=condition, y=metric, order=order,
size=3, ax=ax, hue=hue, hue_order=hue_order, color='.3', dodge=False) #palette=colors,
ax.legend_.remove()
ax.set_ylim(0,1)
ax.set_xlabel('')
ax.set_title(title)
sns.despine()
images¶
In [412]:
images
Out[412]:
In [423]:
st.keys()
Out[423]:
occlusion¶
In [460]:
np.sort(st[st.session_block_name=='occlusion'].fraction_occlusion.unique())
Out[460]:
In [458]:
st.keys()
Out[458]:
In [461]:
st = dataset.stimulus_table
st = st[st.session_block_name=='occlusion']
occlusion_images = st[st.session_block_name=='occlusion'].image_id.unique()
occlusion_fractions = np.sort(st[st.session_block_name=='occlusion'].fraction_occlusion.unique())
image_dict = {}
for image_id in occlusion_images:
image_dict[image_id] = {}
for fraction_occlusion in occlusion_fractions:
file_name = st[(st.image_id==image_id)&(st.fraction_occlusion==fraction_occlusion)].iloc[0].data_file_name
file_index = st[(st.image_id==image_id)&(st.fraction_occlusion==fraction_occlusion)].iloc[0].data_file_index
print(file_name, file_index)
image_data = np.load(file_name)
print(image_data.shape)
image_dict[image_id,fraction_occlusion] = image_data[file_index, :,:]
In [479]:
figsize = (15,15)
fig, ax = plt.subplots(10,6,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
i=0
for y, image_id in enumerate(occlusion_images):
for fraction_occlusion in occlusion_fractions:
ax[i].axes.get_xaxis().set_ticks([])
ax[i].axes.get_yaxis().set_ticks([])
ax[i].imshow(image_dict[image_id, fraction_occlusion], cmap='gray')
if y == 0:
ax[i].set_title(str(fraction_occlusion))
# ax[i].axis('off')
ax[i].set_xticklabels('')
ax[i].set_yticklabels('')
if fraction_occlusion == 0:
ax[i].set_ylabel(int(image_id))
i+=1
fig.tight_layout()
In [ ]:
In [ ]:
oddball¶
In [428]:
st = dataset.stimulus_table
st = st[st.session_block_name=='oddball']
image_dict = {}
for image_id in images:
file_name = st[st.image_id==image_id].iloc[0].data_file_name
file_index = st[st.image_id==image_id].iloc[0].data_file_index
print(file_name, file_index)
image_data = np.load(file_name)
print(image_data.shape)
image_dict[image_id] = image_data[file_index, :,:]
In [429]:
image_dict[112].shape
Out[429]:
In [433]:
plt.imshow(image_dict[112], cmap='gray')
plt.axis('off')
Out[433]:
In [434]:
figsize = (14,5)
fig, ax = plt.subplots(2,7,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
for i,image_id in enumerate(images):
ax[i].imshow(image_dict[image_id], cmap='gray')
ax[i].set_title(str(int(image_id)))
ax[i].axis('off')
fig.tight_layout()
In [436]:
analysis.get_sequence_images()[:2]
Out[436]:
In [439]:
figsize = (10,20)
fig, ax = plt.subplots(11,4,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
i=0
for x, image_id in enumerate(analysis.get_sequence_images()):
ax[i].imshow(image_dict[image_id], cmap='gray')
ax[i].set_title(str(int(image_id)))
ax[i].axis('off')
i+=1
for x,oddball_id in enumerate(analysis.get_oddball_images()):
for x, image_id in enumerate(analysis.get_sequence_images()[:3]):
ax[i].imshow(image_dict[image_id], cmap='gray')
ax[i].set_title(str(int(image_id)))
ax[i].axis('off')
i+=1
ax[i].imshow(image_dict[oddball_id], cmap='gray')
ax[i].set_title(str(int(oddball_id)))
ax[i].axis('off')
i+=1
fig.tight_layout()
transition control¶
In [444]:
analysis.get_sequence_images()[2]
Out[444]:
In [446]:
analysis.get_sequence_images()[:1]
Out[446]:
In [450]:
figsize = (5,40)
fig, ax = plt.subplots(24,2,figsize=figsize, sharex=True, sharey=True)
ax = ax.ravel()
i=0
for x, image_id in enumerate(analysis.get_sequence_images()[:2]):
ax[i].imshow(image_dict[image_id], cmap='gray')
ax[i].set_title(str(int(image_id)))
ax[i].axis('off')
i+=1
for x, image_id in enumerate(analysis.get_sequence_images()[1:3]):
ax[i].imshow(image_dict[image_id], cmap='gray')
ax[i].set_title(str(int(image_id)))
ax[i].axis('off')
i+=1
for x, image_id in enumerate(analysis.get_sequence_images()[2:]):
ax[i].imshow(image_dict[image_id], cmap='gray')
ax[i].set_title(str(int(image_id)))
ax[i].axis('off')
i+=1
for x,oddball_id in enumerate(analysis.get_oddball_images()):
image_id = analysis.get_sequence_images()[2]
ax[i].imshow(image_dict[image_id], cmap='gray')
ax[i].set_title(str(int(image_id)))
ax[i].axis('off')
i+=1
ax[i].imshow(image_dict[oddball_id], cmap='gray')
ax[i].set_title(str(int(oddball_id)))
ax[i].axis('off')
i+=1
for x,oddball_id in enumerate(analysis.get_oddball_images()):
ax[i].imshow(image_dict[oddball_id], cmap='gray')
ax[i].set_title(str(int(oddball_id)))
ax[i].axis('off')
i+=1
image_id = analysis.get_sequence_images()[0]
ax[i].imshow(image_dict[image_id], cmap='gray')
ax[i].set_title(str(int(image_id)))
ax[i].axis('off')
i+=1
fig.tight_layout()
In [ ]: